Notes on
The Extended Phenotype
by Richard Dawkins
I would like to achieve a reasonably objective analysis of this book. Dawkins does raise my hackles however, so I don't think I am anywhere near that at present.
Dawkins presents his thesis as a difference in perspective rather than of substance.
Dawkins argues for gene selectionism, against(?) gene determinism.
This is a chapter in which Dawkins hedges the claim that Darwinian adaptation yields some kind of "perfection".
1. Necker Cubes and Buffaloes
Dawkins presents his thesis as a difference in perspective rather than of substance.
Methodological Discussions
An Alternative Perspective

In "The Selfish Gene" Dawkins tells us that the central thesis of his book is a mere tautology, and I think, that the rest follows logically from it. However, he then goes on to claim that, except in special and limited circumstances this will result in people being selfish.

In "The Extended Phenotype" he does something rather similar (in logical terms). He claims that what he intends to do is not to articulate a scientific theory, but rather to advocate a way of looking at things, which does not involve a substantive empirical thesis. (this is what the Necker cube is there to illustrate, that there can be equally valid ways of thinking about or talking about exactly the same phenomena).

That is "the point" of this first chapter, but before he gets through this chapter he is firmly (perhaps even dogmatically) asserting a point of view on a substantive issue.

The First Paragraph

The first paragraph of this chapter can bear close scrutiny. In it he declares that he is engaged in advocacy, and excuses himself on the grounds that what he is advocating is:

  1. "not a new theory"
  2. "not a hypothesis which can be falsified or verified"
  3. "not a model which can be judged by its predictions"
If it were either of these Dawkins would agree with Wilson (one of his critics) that the "advocacy method" would be inappropriate (and reprehensible). Dawkins claims its OK for him to use the "advocacy method" because he is not doing the things above. What he is advocating is "a way of looking at familiar facts and ideas".

For me there are two principle points of interest raised by this position.

The first is the question of how one can do worthwhile and useful scientific work if not by doing one of the things which Dawkins says he is not doing. I am not here doubting that that is the case, I am genuinely interested in understanding how it can be done, and in understanding how one can tell that some insight presented in this kind of way is true or worthwhile.

A second question to be born in mind while reading the book is whether Dawkins is as good as his word. Does his advocacy extend only to the point of view, or is it also applied to some of the "facts and ideas" of which he presents his novel perspective?

In philosophical writings it seems to me that a substantial part of the content is not in conclusions drawn or stated in the work, but in the presumptions which the author takes, implicitly or explicitly, as given. In this book there is a great deal of groundwork which Dawkins undertakes before presenting his novel perspective, these form the "familiar facts and ideas" on which he will present a new perspective. The earlier parts of this are close to being direct responses to critics of his previous "adversarial" volume (The Selfish Gene [Dawkins89]) and the rest (amounting to the larger part of the book), background of similar material which Dawkins presents as factual background to his ideas on the extended phenotype.

The Necker Cube

The Necker Cube is a line drawing of a cube in which the perspective is ambiguous, so that the interpretation put on the line drawing, in relation to which corners are in the foreground and which in the background, changes between two equally plausible interpretations.

In fact (though this is not noted by Dawkins) this is the case because the perspective is suppressed and the lines in the diagram come in three groups of strictly parallel lines. If we took a photograph, or a realistic painting of a wire cube the lines receding away from the eye would not be strictly parallel, and their convergence or divergence would disambiguate the perspective. Binocular vision provides a further stereoscopic source of disambiguation.

This isn't really significant. If the cube were far enough away, viewed telescopically, the clues can be made as small as we like.

There is nevertheless some point in watching closely to see whether the two perspectives which Dawkins considers could both be accurate accounts of a single underlying phenomenon.

Dawkins' Evolutionary Perspective

The change of perspective which Dawkins advocates is from the supposedly then orthodox view that it is the individual organism which is the beneficiary of evolutionary adaptation, to one in which the beneficiary is the extended phenotype.

Other Matters

While Dawkins is telling us that he offers only a change of perspective he seems to contradict himself quite forcefully. The alternative perspective is only properly to appear in the last four chapters of the book, and quite a bit of preparation is necessary. First Dawkins has to defend himself on "gene determinism" and "adaptationism". There is then to be considerable material devoted broadly to the idea of the gene as the unit of replication and selection and ancillary issues. This displaces the individual organism from these roles (which is apparently where Darwin had them). Then the extended phenotype displaces the individual organism as the locus of expression of genetic material.

It sounds like the main material of the book assumes that the issue on "unit of selection" (which he does mention in this chapter) is just between the organism and the gene, larger units having been ruled out.

Philosophical Connections

Putting aside the question of whether the whole of the work is as Dawkins characterises it (a perspective rather than a substantive claim) and considering only that part (supposing there to be such a part) which does have that character, it is worth considering whether this is Biology or Philosophy, and if a philosopher were to be undertaking such an enterprise, what kind of philosophy it would constitute.

I have tended to think of the non-empirical aspects of science as being either theoretical science which is typically a mathematical exercise in elaborating the consequences of some empirical hypothesis, or as a kind of metaphysics probing beneath the predictive model for evidence of objective reality. Does the non-empirical part of this volume fit either of these, or is it something else altogether.

When Dawkins give his own account of the enlightenment which he derives from the perspective of "the extended phenotype", he talks in terms which are familiar to me from my own supposed insights into human nature arising from readings about evolution:

"I have found that the viewpoint represented by the label `extended phenotype' has made me see animals and their behaviours differently, and I think I understand them better for it."
I experience these kinds of feeling too (not in respect of the extended phenotype, as yet). But sometimes also sometimes see these things in others and doubt whether their sense of enlightenment has any substance behind it. My inclination is to thing such feelings places to start from, as a basis from which to develop something less ethereal, something of which the benefits can be shown to be more tangible.

Dawkins offers a more specific analogue in Biologists' discussions of the possibility of there being more (or less) than two sexes. This he compares with thought experiments in the Philosophy of Mind. At greater length he discusses the use of a "thought experiment" to show that "anything could, in principle, evolve from anything else". (I believe that the argument here connects with an argument which Dawkins presents later to establish that every behaviour has a genetic basis).

Dawkins tells us that thought experiments will be frequently used in the book.

Individual Organisms

Dawkins "thesis" (which is what he calls it here), is that it is legitimate to talk of adaptations as being for the benefit of some kind of entity, but that entity is not the individual organism but the "germ-line replicator" the most important examples of which are genes. (this sounds like a replay of the thesis of "The Selfish Gene").

Replicators are selected by their phenotypic effects, but these should not be construed exclusively in terms of the individual organism. The replicator should be thought of as having an extended phenotype, which encompasses all the effects of the replicator on the world at large, and selection takes place through these effects.

Dawkins digresses briefly to dismiss selection operating at any level higher than the individual organism ("group selection"), so these are not for him legitimate alternative perspectives. That leaves us with Dawkins' analogue of Necker's cube consisting in the two perspectives:

  • for the benefit of the organism, selection on the organism (which is a position he attributes to Darwin)
  • for the benefit of the germ-line replicator, selection on the extended phenotype

2. Genetic Determinism and Gene Selectionism
Dawkins argues for gene selectionism, against(?) gene determinism.
Introductory Observations

I wasn't intending to make notes on my first pass through this book but feel compelled to do so, I'm pretty confident that I don't have this right, and perhaps will understand Dawkins better on the next read, but for future reference (should I ever get to the second pass) I am noting here how this looks to me right now.

Dawkins' Definitions

Curiously the two terms in the title of this chapter do not appear in the glossary. This is what I think he means by these terms.

By genetic determinism Dawkins seems to mean the idea that behaviour (in general, or of some specific kinds) is completely determined by genes.

By gene selectionism he seems to mean the idea that adaptations which have been evolved by natural selection must have a genetic basis. This he seems to treat almost as if it were a tautology, as if it were built into the definition of adaptation, but it doesn't seem to be.

Alternative Conceptions of Genetic Determinism

Now I have read some of Gould, and I am under the impression that Gould means something completely different by genetic determinism, the kind of thing that his essay on Canning's left buttock was arguing against. Furthermore, from reading this chapter I have a sense of why someone might consider Dawkins, despite his protestations, a "genetic determinist" in stronger ways than he seems to acknowledge.

Gould's Conception

I am not sure that he called it "genetic determinism" but here is a kind of determinism which I understand Gould to have been arguing against, for example in his essay on Canning's left buttock Gould91.

The idea is that the characteristics of species are completely determined by the effects of evolution by natural selection (on the gene pool). Gould's belief is that there is a significant admixture of sheer chance, so that life might have been different for really quite absurd reasons.

Gene Selectionism as Genetic Determinism

Given the great distance between Dawkins' definition of genetic determinism and anything which I recall from my readings of Gould, one must also consider the possibility that others who have criticised Dawkins as a genetic determinist have also had in mind a concept widely different from the one which he has defined. From reading his own story about interchanges at conferences it seems to me entirely possible that some of his detractors might mean by "genetic determinism" just what Dawkins defines as "gene selectionism".

This is because Dawkins seems to take it as obvious that any adaptation which has evolved by natural selection must have a genetic basis, and seems to regard it important to call a spade a spade. So when others talk about adaptations without referring to genes, keeping an open mind about whether the adaptation has a genetic or perhaps a cultural basis, Dawkins will enter into the discussion and talk about the evolution in terms of the action of natural selection on genes. He thinks he is doing no more than making explicit what was implicit in their talk, but they think he is indulging in a prejudice about how the characteristic is transmitted (which they might possibly call "genetic determinism").

For and Against Gene Selectionism
Dawkins For

Dawkins has an argument by which he supports the idea that adaptations always have a genetic basis. It is along the lines that it always will be possible to find some genetic variation (not necessarily one which has survived in the gene pool but one which can be found in previous stages of evolution) which makes the behaviour under consideration impossible. If you go back far enough you come to bacteria and there probably are no behaviours of mammals which can be exhibited by bacteria. Once you have found this disabling genetic variation, then its inverse variation must qualify as a genetic basis (if perhaps a redundant one) for the behaviour.

Me Against

I don't know if this is the basis for Dawkins apparent belief in gene selectionism, but it does seem to me to be fallacious. For a genetic variation to constitute the genetic basis for some behaviour trait I think one would need stronger conditions to obtain. It is the case that all behaviours have a genetic basis in the sense of their being some conditions on the genetic structure which are prerequisite for that behaviour. But the evolution of the behaviour might take place much later than the evolution of its minimal genetic prerequisites, so that evolution would not properly be described as resulting from the effects of natural selection on the propagation of genes. The best explanation might be in terms of the effects of natural selection on behaviour patterns taught by parents to their children.

More Terminological Issues

On my second reading of this chapter, some further points stand out.


According to the glossary `an attribute of an organism which is "Good" for something'

Darwinian Adaptation

This is not in the glossary, but Dawkins says:

"The existence of a Darwinian adaptation, then, implies the some time existence of genes for producing the adaptation"
So we may wonder, is this just part of Dawkins' definition of Darwinian adaptation, or does he hold that all adaptations have a genetic basis.

Natural Selection
(not in glossary)
"It is always possible to talk about natural selection of a behaviour pattern in two ways"
(the first is in terms of individuals exhibiting the behaviour being "fitter")
"Alternatively, and equivalently, we can talk of genes for performing the behaviour pattern surviving better than their alleles."
Darwinian Selection
"It is always legitimate to postulate genes in any discussion of Darwinian adaptation ... Objections ... betray a fundamental failure to face up to what Darwinian adaptation is about."

The possibility that Dawkins appear not to entertain, but to which others seem open, it that evolution by natural selection might take place on characteristics which are transmitted by other than genetic mechanisms, e.g. transmitted from parents to children by example.

When accused of being a genetic determinist, in the examples he himself cites, this seems to be precisely because he presumes that evolution must always be of genetically transmitted characteristics.

This seems to baffle Dawkins, possibly for the following two reasons. Firstly because the possibility his adversaries wish to admit Dawkins considers a mere confusion. Secondly because his own conception of genetic determinism is more radical, than that of which he is accused, and he cannot understand why he is accused of this radical fallacy.

3. Constraints on Perfection
This is a chapter in which Dawkins hedges the claim that Darwinian adaptation yields some kind of "perfection".
Some Terminology

The title is "constraints on perfection" so Dawkins is defending the claim that evolution yields some kind of optimal adaptation over the long run by admitting limited special circumstances in which the result of evolution appears not to be quite "perfect".

up quick index © RBJ

privacy policy


$Id: dawkins1982.xml,v 1.5 2010/01/10 21:30:11 rbj Exp $